Dean Jim Morris's Thought of the Week

Published 09.29.06

SaaS SaaS is pronounced as “sass.” Not, as you might think from the spelling, “sauce.” “SaSS” might be a better acronym. Rather than “Software as a Service”, it could mean “Service as the Savior of Software.” The combination of the bust, open source, and off-shoring has made the software community glum. Larry Ellison declared it a goner and the president of Oracle Online, Tim Chou, wrote a book called The End of Software. Now comes the SaaS movement, formerly called the ASP business, complete with its own conference, SaaSCon, keynoted by none other than Tim Chou and attended by several of the people for whom you think SaaS was not good – Microsoft, IBM, and many enterprise software consultants. Notably absent were many of the companies that are making a bundle from service – Google, eBay, Amazon, etc. Let’s be clear – a web service solution that replaces a licensed product is going to be much cheaper, so this movement is really about reducing software delivery costs, just as Nicholas Carr suggested in “IT Doesn’t Matter.” There may be wonderful, innovative things coming later, but SaaS is really a cover story for a significant retrenchment of the software industry. posted - 9/29/2006 Where is the Web Terminal Company? There have been many past attempts to create and sell the “Network Terminal,” most notably by Larry Ellison in the 1990’s. Now that improvements in broadband, Ajax, and the many SaaS applications are rolling out, selling a box that does nothing but browse the Web seems appropriate. My spec for such a machine would be a conventional PC with as much disk as I can afford, but it all belongs to the system for caching Web pages. The software would be Linux and Firefox. I can keep any private data on a memory stick, but mostly it would be kept on the Web somewhere. While there is no savings in hardware costs, there are savings in software as it is likely to be free on the web. The biggest savings is in system administrator time. The terminal would be saving users from themselves by making it impossible to install and upgrade software packages on their own. In today’s environment, that’s what ends up costing far more than the hardware. There must be several people doing this in stealth mode. posted - 9/21/2006 Is the Bay Area Good for Entrepreneurs? “Duh,” you say? Well, according to Entrepreneur.com( http://www.entrepreneur.com/bestcities/region/large.html), we rank 28th, after such hotbeds of innovation as Columbus, Ohio! There is something suspicious about this ranking, but I’m beginning to wonder whether the Bay Area is, in fact, the best place to grow a business. Housing costs are a major issue along with the general expense of doing business. The CEO of a start-up that recently moved here from the heartland said that his team was more focused on creating the product and less tempted to job hop in its previous location. Maybe the right approach is to start the company here with the founders, keep the headquarters here for deal making, but do all the development work elsewhere. You don’t have to go all the way to Bangalore; Pittsburgh would do nicely. Maybe the Bay Area is the new Boston – a place where you come to learn and get your ticket punched, but too expensive to create a business. posted - 9/8/2006 Software Product Management and the Endless Beta When software delivers its service over the web, we can do business differently. Our ability to control our software’s environment—all of it that runs on our servers, anyway—is helpful. The ability to fix the software without distributing updates is helpful. Since interactions between users and servers go over the net, they can be recorded and replayed. All of these things can be exploited to support much better bug analysis and performance monitoring. Software just got easier! On the other hand, being able to monitor users is complicated by their huge numbers on the net. What do you do with millions of traces of interactions? To begin with, the application should be seeded with exception conditions for which the programmer would like to receive a report. These reports can catalogue bugs or any other condition the writer considers noteworthy; e.g. a particular feature has been used in an unexpected way. Performance data should be aggregated. An application should evolve to better serve its purpose. So, aside from providing a service that attracts users, it should be gathering information about what else the users might want or need. Thus, the tough decisions that product managers make about which features to include in the next release can be made in a more informed way. The product management team implements features in a rudimentary form and then tracks what the users do with them. In this model, features can be tried on small subsets of users until they prove dependable and popular. posted - 8/30/2006 Silicon Valley 101 This week, I have a question. Many people come to Silicon Valley to learn the magic of entrepreneurship and innovation. If we hosted a person here for a period of time, what could we do to maximize his or her learning and the likelihood that he or she would absorb the “magic”?

• Have the students work in a start-up
• Have successful entrepreneurs speak
• Have venture capitalists speak
• Listen to a business plan pitch
• Have students attend public seminars such as the:
  • SD Forum
  • Churchill Club
  • Computer History Museum
  • PARC
• Take a course in entrepreneurship from a well-respected university
• Read Tom Byer’s book, Technology Ventures: From Idea to Enterprise
• Learn from fellow students
• Provide company visits
• Provide networking opportunities with individuals working at small companies

How would we select the best applicants to the program?

• Previous start-up experience
• Academic performance
• A written statement
• A proposal for a new enterprise
• Extracurricular activities in college

How short or long should such a course be? Can a simulated start-up experience be created that is as effective as a real one? posted - 8/21/2006 A Robot Has A Ball It's too hot to think this week. Here is a cool video about the most recent Carnegie Mellon robot. http://video.google.com/videoplay?docid=9117167457978282229 The video introduces a new type of mobile robot that balances on a ball rather than on legs or wheels. "Ballbot" is a self-contained, battery-operated, omnidirectional robot that balances dynamically on a single urethane-coated metal sphere. It weighs 95 pounds and is the approximate height and width of a person. Because of its long, thin shape and ability to maneuver in tight spaces, it has the potential to function more effectively than current robots can in environments with people. Ballbot's creator, Robotics Research Professor Ralph Hollis, says the robot represents a new paradigm in mobile robotics. What began as a concept in his home workshop has been funded for the past two years with grants from the National Science Foundation. Hollis is working to prove that dynamically stable robots like Ballbot can outperform their static counterparts. Traditional, statically stable mobile robots have three or more wheels for support, but their bases are generally too wide to move easily among people and furniture. They can also tip over if they move too fast or operate on a slope. Ballbot has an onboard computer that reads balance information from its internal sensors, activating rollers that mobilize the ball on which it moves - a system that is essentially an inverse mouse-ball drive. When Ballbot is not in operation, it stands in place on three retractable legs. Hollis noted that current legged robots, such as humanoids, are complex and expensive. He's looking for simple alternatives to better understand the issues of dynamic stability for mobile robots in human environments. He believes that the research may produce a robot that could have useful, meaningful interactions with people who are elderly, disabled or need assistance in an office environment. Hollis and his team - including Robotics Institute Project Scientist George Kantor and graduate students Tom Lauwers, Anish Mampetta and Eric Schearer - have demonstrated Ballbot moving on carpeted surfaces. They presented their research findings in October, 2005 at the prestigious International Symposium for Robotics Research in San Francisco and, most recently, at the International Conference on Robotics and Automation, which took place in May in Orlando, Fla. Future plans for Ballbot include adding a head and a pair of arms. Swinging the arms, said Hollis, would help to rotate and balance the body. "We want to make Ballbot much faster, more dynamic and graceful," he said. "But there are many hurdles to overcome, like responding to unplanned contact with its surroundings, planning motion in cluttered spaces and safety issues." Hollis has been a pioneer in the field of mobile robots since he began building them as a hobby in the 1950s - well before there were commodity transistors, personal computers or easily accessible off-the-shelf parts. In the 1960s, he developed one of the world's first mobile robots and followed that in the 1970s with the Newt mobile robot, which was one of the first to have an onboard computer. Hollis wrote an article about Newt for the now-defunct Byte Magazine that was voted one of the publication's best stories of all time. Newt subsequently became a subject in the NOVA television documentary "The Mind Machines." posted - 8/14/2006 Can Computers Generate A Next Big Thing? Today's opinion is that the next big thing will not be in computers. Yes, the Internet is still growing, but venture capitalists want to invest in things whose future growth has not already been recognized by the market. As a result, investment, excitement, and the people who follow it are migrating to biotech, alternative energy, and other things not computing. However, we underestimate the potential of the computer business because of a blind spot about non-linear growth. We tend to predict the future using a linear model; e.g. if sales grew by x units this year it will grow by 2x in two years. Some markets have quadratic growth—if there are x Internet users today, there will be x2 next year. Exponential growth, represented by Moore's law, is more ferocious—chip performance of x going to 2x each year means that chips improve as much each year as in all previous years combined. Each year or two could bring a paradigm shift if other things did not dampen the effect. What can dampen growth? There are fundamental limits like the total number of people in the world, so the number of human Internet users can't grow forever. Another dampener is human choice represented by investment and government policy. For example, in the 15th century, the Chinese had a flourishing oceanic exploration program, but the government decided to kill it, which created a lucrative opportunity for the Europeans. "Do the math" often means deciding which of two large numbers is bigger. If investors did real math that analyzed non-linear growth, they might make different decisions. posted - 7/14/2006 Who Will Be The Network Police? It’s common to describe an operating system as a traffic cop that makes programs share resources fairly. Historically, the resource to be managed was memory. However, in the age of increasingly less expensive memory, we might not need a policeman for that, but we surely do for other things. Screen space remains a scarce resource, not just because of screen size but also because human eyeballs can only take in so much. The creators of software are motivated to dominate screen space to capture the brains behind those eyeballs. I once fantasized that Windows might become a screen-space policeman if Microsoft were broken up and the OS part were declared a regulated monopoly. In return for being a sanctioned monopoly, Microsoft would serve as a screen-space policeman. As it turned out, Microsoft kept favoring its own applications in the screen space wars. Now the closest thing to a good traffic cop is Google, which restricts advertisers to the right side of the screen. Google was the first to discover we didn’t like being accosted by banners and pop-ups. The need for policing goes much further. As science fiction writer Bruce Sterling observes, cyberspace has become a slum—full of scammers and spies. Now real lawmakers and police are being called in, but they are outmatched by the dark element of the hacker community. Maybe Microsoft, like the legendary gunslinger, will sign up to be the sheriff now that it is being out-gunned by Google. posted - 7/6/2006 Genius Trumps Literacy Paul McCartney, who turned 64 this week, can’t read or write sheet music. When he wrote a symphony, Standing Stone, he used a computer to translate his sounds into music notation: He played a keyboard connected to a computer that would then transcribe the music he played into sheet music. Muhammand was illiterate. So was Joseph Smith, the founder of Mormonism. Finally, I’m sure Captain Kirk of the Starship Enterprise can’t read. I’ve watched every episode, and he’s never read anything. He just asks questions and is given answers by Mr. Spock, who does read. posted - 6/19/2006 Silicon Valley is Different When I look at Silicon Valley with my Pittsburgher’s eyes, several things strike me. It’s in California, but it’s not laid back. People here are overworked. Engineers and business men and women work 60-hour weeks, and they are the lucky ones. Service workers and others without high-tech skills work 80-hour weeks on multiple jobs. People here might seem friendlier than Easterners, but it’s the way flight attendants are friendly until the flight is over. Everyone is polite and pleasant, but it’s all business. If you try to get away from it with a leisurely bicycle ride, you’re run off the road by packs of whippet-thin cyclists on $5,000 bikes wearing all spandex. In most places, changing jobs frequently is bad for your resume. In Silicon Valley it’s good. If you’ve been working at the same company for ten years, it shows either that you are not a risk taker or that your co-workers, who left the company for a start-up, didn’t take you along. Ann Saxenian believes that Silicon Valley thrives because its workers and investors have no company loyalty. Their loyalty is to the “next big thing.” When that next big thing was the PC, they abandoned companies like Hewlett-Packard, which was making mini-computers, and they went to Apple Computer. In contrast, Boston was dominated by Digital Equipment and Ken Olson, who didn’t believe in the PC. Silicon Valley is not driven by visions. New ideas and visions are needed, but they are plentiful everywhere. The businesses of Silicon Valley are driven by people who have discipline and focus. They live by the motto “The main thing is to keep the main thing the main thing.” posted - 6/12/2006 The Endless Beta Google News spent three years in Beta release, GMail is still in Beta, and the Google Earth T-shirt notes that the Earth itself has been in Beta for 4.7 billion years. I’m beginning to get the message: In the future, many software offerings will have a minimal Alpha phase before release to the public and no Gamma phase (if that’s what you call the period after Beta). The development team will get the software up and running, release it, and keep enhancing it for its lifetime, perhaps changing the software daily. Software wants to change; it’s the nature of the beast. All hackers wish to operate this way – and it takes years of training, some of it painful, before they accept the rigorous test and release regime of industrial software. The Software as a Service business model seems to offer us an escape back to the time when every bug could be fixed immediately and every new idea tried instantly. However, the Endless Beta requires a new kind of software engineering. The freedom to release at will must be tempered with the discipline to make every release an improvement with fewer bugs and better features. The practices of the Open Source community with its hierarchy of release controllers (a.k.a. “committers”) demonstrate one successful method: Each person’s changes are checked by a more responsible person in the community. The method for testing will change since the entire user community will be an involuntary test organization. The software will contain elaborate monitoring facilities that log every bug and difficulty users encounter without their awareness. The development team will come in every morning (or night), apply triage to the logs, address the most important things, and put out a new release. The pace of testing and fixing will become relentless with competitors struggling to match each others’ offerings on a daily basis. It will seem like an endless programming contest. Be careful what you wish for, hackers. posted - 6/2/2006 Teleportation, Time Travel, and Immortality With Moore's law still in effect, we need some ambitions with exponential growth. Science fiction and religion express some of our deepest desires, so let's talk about the prospects for teleportation, time travel, and immortality. Our technology won't be able to deliver these in quite the way you want, but we can approximate them in just the way today's aircraft achieved Icarus's dream of flying with his own muscles. For example, we can provide teleportation, not by moving you, but by moving the situation to you. Someone recently performed heart surgery remotely (see http://www.military.com/features/0,15240,97023,00.html for a recent report). The same technique also works for time travel as long as you're willing to restrict your travels to the past. Using image capture and analysis, we can capture a complete event in three dimensions with sound and smells. Then we can make a virtual reality Karaoke machine that allows you to insert yourself into the action (see http://www.ri.cmu.edu/labs/lab_62.html for Takeo Kanade's project, which will allow you to play the role of the referee in a past football game). Immortality, however, will take a few more years to accomplish. We start by capturing everything that happens in your life. Gordon Bell is working on this (see http://portal.acm.org/citation.cfm?id=381893&dl=ACM&coll=ACM). Then a suitable amount of machine learning should be able to create the algorithms you use for all behavior. These algorithms are then downloaded to a robot that looks like you. If this doesn't meet your definition of immortality you might buy the real thing from Ray Kurzweil at http://www.rayandterry.com/wellness_one.php?osCsid=f9443e14b6be787b298a2e5b7fe59e7b posted - 5/30/2006 Is A.I. Crap? At a meeting of leading computer scientists, a presentation by one was followed by another asking, “Isn’t this just the old Artificial Intelligence crap?” Aside from showing that rudeness is alive and well in our field, the question is useful, and its premise should be addressed. Indeed, the announced purpose of the meeting was to breathe some life into an allegedly intellectually timid computer systems community, and A.I. might do the trick. Sometimes I look at academics as a sort of intellectual digestive system, taking in new knowledge and ideas and incrementally processing them to nourish society with new capabilities and graduates. In this model, the entry to the computer science digestive system is manned by the science fiction writers like Neal Stephenson and George Lucas. They examine the technological landscape, look at humankind’s needs and aspirations, and then tell a story about how it might all come together. (Jules Verne should get at least a little credit for NASA’s trips to the moon.) Next in line along the intellectual tract is MIT’s Media Lab, which has often tried new ideas and stunts that others apply years later. At this point, scientists in artificial intelligence take these ideas and create prototypes that are more widely applicable. Next along the intellectual digestive tract come various systems engineers who perfect the prototype and create systems that are engineered and understood. Finally, the theoreticians reduce ideas and questions to their simplest essence, putting them in a form that can be easily taught and remembered. Each stage in the digestive system has an important role to play and where one decides to work is more a matter of temperament than talent. As to which end produces crap, I’ll let you decide. posted - 5/12/2006 Forward to the Past Remember 1984? The PC was shipping, the Macintosh was introduced, AT&T was about to be broken up, and our computing model was timesharing. Technology has changed a lot since then, but we're about to go back to a centralized computing model enabled by the Internet. The PC is about to become a high-powered AJAX terminal. And AT&T is back, too. Before you shudder, realize that AT&T created our model for what a good computer-based service should be: a large, complex infrastructure supporting a very simple interface to a vital service that is ubiquitous and always up. Let's hope the Internet becomes as serviceable as the old black telephone. The future belongs to services transmitted by the Internet and delivered on many different kinds of devices. posted - 5/5/2006 Hire Lobbyists Three groups of interested parties threaten the growth and health of the Internet. The first group is the communications companies such as AT&T, Verizon, etc., which are fighting to control access and service levels on the Internet. The second group is the content providers such as publishing (including print, music, and video) companies and database providers who will fight to prevent universal access and fair use copying. And the third group is governments, which may resist, restrict, and/or monitor information exchange, as is already happening in China. All three of these groups have massive lobbying operations, primarily in Washington D.C. But the beneficiaries of the Internet remaining free, open, and neutral – Google, eBay, Amazon, et al. – have very few lobbyists. The battle over control of the nations’ communications infrastructure has been going on for over a century, and new companies must get in the game. posted - 4/28/2006 Sell to the Long Tail Chris Anderson1, the editor of Wired magazine, observes that an Internet bookstore like Amazon.com can stock 20 times as many titles as a physical book store. A graph showing total sales of all titles ranked by quantity would exhibit a very long curve approaching zero, its long tail. While many of those titles are sold in small numbers, they collectively comprise 20% of Amazon’s revenues. The same is true for sellers of movies and music. The Internet’s lower selling costs allow more product diversity, larger customer bases, and better target audiences for the advertisers. [1] http://www.wired.com/wired/archive/12.10/tail.html posted - 4/21/2006 Don’t Compete with Microsoft™ The search for new software business models is partly based on the need to avoid Microsoft. The greatest fear of a software product developer is that Microsoft will offer your product for free, and venture capitalists will often ask entrepreneurs how they intend to compete with Microsoft. However, not competing with Microsoft is easier said than done - Microsoft took away Netscape’s oxygen by offering a competing product as a component of a bundled product. posted - 4/14/2006 Use Advertising to Make Money If you are thinking of a business based on a web service, Goggle and Yahoo! offer you a business model that simplifies your life. Give the service for free but, to generate revenue, provide relevant advertising links. You eliminate the worries about pricing, collecting for services, and other business details, allowing you to focus on what’s really critical – creating a service that attracts a lot of traffic, which is the hardest part of starting a business. Sales resistance from pricing is minimized because your service is free, and users are more forgiving if your service doesn’t work perfectly in the beginning. Doing a start-up just got easier. posted - 4/7/2006 Serve Software Engineers Clayton Christiansen suggests exploiting intense competition at points on the value chain by working upstream or downstream from those points of high competition. The downstream play is more obvious – competitive forces should result in lower prices, so one should be a buyer. The upstream play is what arms merchants do – they sell tools to the guys who are desperate to gain an edge. The global oversupply of software creators suggests there is a market for things that software producers need to gain a competitive edge. Four examples include:

Slashdot is a very successful web service whose motto is “News for nerds, stuff that matters.”

SourceForge is a clearing house for open source software components.

In some quarters, a high rating in the Capability Maturity Model is essential to compete, so teaching it is a good business.

Software production tools that convey a unique advantage might also be a viable business if the tools cannot be displaced by open source versions. posted - 3/31/2006 Use Open Source Software. In the earlier days of the computer business, computer companies gave their software away to encourage hardware sales. As time went by, companies realized that selling software was lucrative as well. However, a group of programmers led by Richard Stallman created a movement that encouraged the free exchange of software. These progressive programmers believed that software and all other information should be free and, ultimately, they created the system now known as GNU-Linux. Many other similar groups have followed resulting in the creation of a common base of software that now serves the world well. Open source software is nominally free, but it has costs. Most organizations buy support services from companies such as Red Hat and Spikesource. These open source support firms not only provide services that enable the various applications to work together, but they also inform companies of updates and other new related software. Perhaps the biggest advantage of open source software is that software engineers can fix or extend the software themselves, if necessary. Another advantage is that because the code can be inspected, it’s an educational tool for programmers. posted - 3/24/2006 Hire Great Software Engineers. In the software business you need the best minds you can get. Software engineers not only know programming but can design products, create highly reliable systems, and control costs. Hire them from a competitor, hire computer scientists and hope they become engineers, or hire Carnegie Mellon graduates who are great software engineers out of the box. posted - 3/17/2006 Sell Service, Not Software Google, eBay, Amazon.com, Yahoo!, and other successful new companies are built on software. Salesforce.com, a sales support enterprise, could have sold software licenses to businesses, but it decided to simply sell access to a web site. The most compelling case for not licensing software stems from the nature of software itself – it’s easy to change and easy to copy. Changing Software. Traditional software vendors release products that can’t be changed once out the door. It may take years for the customer to be able to buy an update of even the smallest bug fix. In the meantime, vendors must support an expensive system of help desks, work-arounds, and patches. In contrast, when a pure service provider discovers a bug, the company can fix it whenever it wants to. The diagnosis and repair of problems are much easier because the service provider controls all the hardware and surrounding software. The company doesn’t have to ask questions such as “Do you have an account?”, “What version?”, and “What environment?” posted - 3/10/2006 Copying Software. The other bugaboo of software is piracy. If you sell someone the code for a product, they can give their friends copies of it. There are huge factories in China that duplicate software. Sellers of software must pursue countermeasures, lawsuits, and negotiations to stem software piracy. The Internet has made the problem much worse by enabling cost-free distribution of any digital material. Selling a service rather than software, however, eliminates both these problems in a single stroke - the company can update software easily because it’s the only one running the software, and there is no piracy because there is no code to copy. Furthermore, keeping software in-house ensures that innovations remain trade secrets. Eric von Hippel and Steven Klepper have shown that manufacturers innovate more in their processes than in the products themselves. If the company’s clever ideas remain inside the business, customers and competitors cannot discover trade secrets through reverse engineering. [1] von Hippel, Eric, The Sources of Innovation, Oxford University Press, 1988. [2] Klepper, Steven, “Entry, exit, growth, and innovation over the product life cycle,” American Economic Review, 1996. posted - 3/3/2006 Ride the Internet wave Now that the Intenet has taken off, it will grow more and more rapidly until every man, woman, child, and dog on earth is connected. A gravitational attraction called Metcalfe's Law encourages people to join in. Economist Brian Arthur[1] predicts the Internet will follow a pattern similar to previous large infrastructure projects, such as ­canals and railroads, in which the infrastructure grows robustly after an investment bubble. The Internet bubble signaled the beginning, not the end. Any business that exploits the growth of Internet use gets a natural boost. Three enemies threaten[2]: (1) The entrenched communications interests will fight to own the net, (2) The content providers will fight to restrict copying, and (3) Governments might resist or monitor information exchange. Let’s hope they all lose. [1] Arthur, W. Brian, “Why Tech Is Still The Future,” Fortune, November, 2003. [2] http://www.truthout.org/docs_2006/020406I.shtml posted - 2/24/2006

VIEW MORE NEWS   
SUBMIT NEWS